IBM Mainframe Forum Index
 
Log In
 
IBM Mainframe Forum Index Mainframe: Search IBM Mainframe Forum: FAQ Register
 

Tuning of Large Sorts using SYNCSORT


IBM Mainframe Forums -> JCL & VSAM
Post new topic   Reply to topic
View previous topic :: View next topic  
Author Message
dhok_ks
Warnings : 1

New User


Joined: 09 Jun 2006
Posts: 1

PostPosted: Wed Jan 03, 2007 1:34 pm
Reply with quote

I want to reduce run the elapsed time as well as the CPU time for the sort step.There are two instances:-


1.Here the input file is of VB and the LRECL is more than 30000 bytes.Also the file has 6 million records.I need to remove the duplicates from this file.This step runs for 4 hours and costs around $70.Please suggest any tuning for this.

2.There are two files which have combined of 88 million records and the record length is 3625.This step runs for 6 hours and costs $60.Please suggest a way for tuning this.

Note:- The estimated cost is seen in the JESYSMSG. I tried to use the VSCORET parm on the EXEC of the sort and the FILSZ parameter in the sort card bu there is not much reduction in the elapsed time.CPU time as well as the cost.

I am using SYNCSORT.

Regards,

Kaustubh
Back to top
View user's profile Send private message
William Thompson

Global Moderator


Joined: 18 Nov 2006
Posts: 3156
Location: Tucson AZ

PostPosted: Wed Jan 03, 2007 2:24 pm
Reply with quote

William Thompson wrote:
Syncsort does deal with large sorts:
Quote:
MAXSORT is a maximum capacity sort designed to sort amounts of data that are too large for an ordinary sorting technique to process.
And long sorts:
Quote:
PARASORT improves elapsed time performance for sorts whose input is a multi-volume tape data set and/or concatenated tape data sets. Reduced elapsed time can help critical sort applications achieve batch window goals.
And also compares tecniques for "Performance Considerations". DO you have a manual?
Back to top
View user's profile Send private message
raak

Active User


Joined: 23 May 2006
Posts: 166
Location: chennai

PostPosted: Thu Jan 04, 2007 12:02 pm
Reply with quote

hi thompson,

In our system also, we have the problem of Extended batch window due to high elapsed time from big sort jobs. But In my system, this PARASORT and MAXSORT is not present.
Is there any other way in which the tape files can be sorted in a much faster way than the usual sort?
Back to top
View user's profile Send private message
William Thompson

Global Moderator


Joined: 18 Nov 2006
Posts: 3156
Location: Tucson AZ

PostPosted: Thu Jan 04, 2007 7:16 pm
Reply with quote

Do you use Syncsort?
What version?
Do you have the Manual?
Back to top
View user's profile Send private message
raak

Active User


Joined: 23 May 2006
Posts: 166
Location: chennai

PostPosted: Thu Jan 04, 2007 7:50 pm
Reply with quote

yes Thompson,

we use SYNCSORT. Version is shown as 3, release 7.

I dont have the manual.
Back to top
View user's profile Send private message
UmeySan

Active Member


Joined: 22 Aug 2006
Posts: 771
Location: Germany

PostPosted: Fri Jan 05, 2007 3:10 pm
Reply with quote

Hi !

A nearly same situation i solved by changing VSCORE and VSCORET to a adaquate size.

Also eliminating not nessesary sort-statements like SUM FIELDS=NONE
This option will load some internal routins although they are not used.

You wrote, your's sorting tape files. Can you load them on DASD first and then do the sort ??? Alloc them on real fast DASD, no multivolume, if possible same special reserved disks, allocations whith a minimum of extents, means mostly full primary allocation.

Also REGION=0M ???

Which JobClass, use a clas with high performance.


Have a look also at the DynAlloc-Parms of your Sort-Installation.
Don't specify Sort-Wk's in JCL.


//STEP02 EXEC PGM=SORT
//SYSOUT DD SYSOUT=*
//*
//$ORTPARM DD DSN=IER1337.MIA01250.SORTSPLT(SORTPARM),
// DISP=SHR

--> $ORTPARM = VSCORE=5M / VSCORET=30M

Regards, Umeysan
Back to top
View user's profile Send private message
raak

Active User


Joined: 23 May 2006
Posts: 166
Location: chennai

PostPosted: Fri Jan 05, 2007 4:23 pm
Reply with quote

Hi UmeyScan

It is not possible to use a DASD in my case. I have so many Tape files and all these contains records in the count of millions. creating large DASD files for all these TAPES just to do the sort is not possible.

Can u provide me some other efficient way in which Tape files can be sorted as they are without putting in to a DASD.
Back to top
View user's profile Send private message
UmeySan

Active Member


Joined: 22 Aug 2006
Posts: 771
Location: Germany

PostPosted: Fri Jan 05, 2007 4:43 pm
Reply with quote

Hi raak !

Does this eventually mean, that you have more than one sortin.
I mean somthing like sortin01 on tape sortin02 on tape and so on ???

If so, sorry, you have to life with a hudghe elapsed time.

Perhaps your shop could send a message to syncsort for some help
of a specialist over there.

Regards, UmeySan
Back to top
View user's profile Send private message
View previous topic :: :: View next topic  
Post new topic   Reply to topic View Bookmarks
All times are GMT + 6 Hours
Forum Index -> JCL & VSAM

 


Similar Topics
Topic Forum Replies
No new posts How to split large record length file... DFSORT/ICETOOL 10
No new posts Compare only first records of the fil... SYNCSORT 7
No new posts Parsing Large JSON file using COBOL COBOL Programming 4
No new posts Split large FB file based on Key coun... DFSORT/ICETOOL 4
No new posts Count Records with a crietaria in a f... DFSORT/ICETOOL 5
Search our Forums:

Back to Top